Building a Synthetic Biomedical Research Article Citation Linkage Corpus

14533

Authors

nan

Abstract

AbstractCitations are frequently used in publications to support the presented results and to demonstrate the previous discoveries while also assisting the reader in following the chronological progression of information through publications. In scientific publications, a citation refers to the referenced document, but it makes no mention of the exact span of text that is being referred to. Connecting the citation to this span of text is called citation linkage. In this paper, to find these citation linkages in biomedical research publications using deep learning, we provide a synthetic silver standard corpus as well as the method to build this corpus. The motivation for building this corpus is to provide a training set for deep learning models that will locate the text spans in a reference article, given a citing statement, based on semantic similarity. This corpus is composed of sentence pairs, where one sentence in each pair is the citing statement and the other one is a candidate cited statement from the referenced paper. The corpus is annotated using an unsupervised sentence embedding method. The effectiveness of this silver standard corpus for training citation linkage models is validated against a human-annotated gold standard corpus.

Keywords

nan

Full Content

Proceedings of the 13th Conference on Language Resources and Evaluation (LREC 2022), pages 5665–5672 Marseille, 20-25 June 2022 © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0 5665 Building a Synthetic Biomedical Research Article Citation Linkage Corpus Sudipta Singha Roy and Robert E. Mercer The University of Western Ontario London, Ontario, Canada ssinghar@uwo.ca, mercer@csd.uwo.ca Abstract Citations are frequently used in publications to support the presented results and to demonstrate the previous discoveries while also assisting the reader in following the chronological progression of information through publications. In scientific publications, a citation refers to the referenced document, but it makes no mention of the exact span of text that is being referred to. Connecting the citation to this span of text is called citation linkage. In this paper, to find these citation linkages in biomedical research publications using deep learning, we provide a synthetic silver standard corpus as well as the method to build this corpus. The motivation for building this corpus is to provide a training set for deep learning models that will locate the text spans in a reference article, given a citing statement, based on semantic similarity. This corpus is composed of sentence pairs, where one sentence in each pair is the citing statement and the other one is a candidate cited statement from the referenced paper. The corpus is annotated using an unsupervised sentence embedding method. The effectiveness of this silver standard corpus for training citation linkage models is validated against a human-annotated gold standard corpus. Keywords: citation linkage, semantic similarity, automatic corpus annotation 1. Introduction There are a variety of formats, writing styles, and pur- poses for different types of written documents. It is possible for a research article to reflect a current trend in the field of study, a new invention, or a novel ap- proach to solving a specific problem. During the pro- cess of writing a research paper, the author examines past studies that are either important in solving the topic at hand or have impacted the author’s current research paper ideas. Using a citation is the process to refer to another article in the current research arti- cle (Houngbo, 2017). In this way, citations serve as bridges between different research papers. Citations free up the authors’ time by removing the need to re- peatedly write the same thing. While doing so, it pro- vides readers with some context for the issues being discussed in the body of the piece. The concept of citation indexing was first introduced in 1964 by Garfield (1972) where indexes contain the en- tirety of the references in a research document. Since then, various analyses of citing have been presented (e.g., (Ritchie et al., 2008)). In biochemistry and physics research papers, Garzone and Mercer (2000) presented a method for determining the objectives of different citations. Furthermore, citation aids in the tracking of logical argumentation throughout multi- ple research articles (Mercer, 2016). Citation is com- monly used to maintain the trail of scientific research argumentation across different scientific papers …

Content truncated to 500 words

Labeled Text

AbstractCitations are frequently used in publications to support the presented results and to demonstrate the previous discoveries while also assisting the reader in following the chronological progression of information through publications. In scientific publications, a citation refers to the referenced document, but it makes no mention of the exact span of text that is being referred to. Connecting the citation to this span of text is called citation linkage. In this paper, to find these citation linkages in biomedical research publications using deep learning, we provide a synthetic silver standard corpus as well as the method to build this corpus. The motivation for building this corpus is to provide a training set for deep learning models that will locate the text spans in a reference article, given a citing statement, based on semantic similarity. This corpus is composed of sentence pairs, where one sentence in each pair is the citing statement and the other one is a candidate cited statement from the referenced paper. The corpus is annotated using an unsupervised sentence embedding method. The effectiveness of this silver standard corpus for training citation linkage models is validated against a human-annotated gold standard corpus.

Publication Details

Paper ID:
14533
Year:
2022
Publication:
TIPSTER
Publication Type:
conf
DOI:
nan